Finally, Paper IV (Edelson et al. 1995) presents a detailed study of the short time-scale variability exhibited by NGC 4151 with emphasis on the multiwavelength ...
arXiv:astro-ph/9605081v2 16 May 1996
MULTIWAVELENGTH OBSERVATIONS OF SHORT TIME-SCALE VARIABILITY IN NGC 4151. III. X-RAY AND GAMMA-RAY OBSERVATIONS
R. S. Warwick1 , D. A. Smith1 , T .Yaqoob2 , R.Edelson3 , W. N. Johnson4 , G. A. Reichert2 , J. Clavel5 , P. Magdziarz6 , B.M. Peterson7 , A. A. Zdziarski8
1 Department
of Physics and Astronomy, University of Leicester, Leicester, LE1 7RH, UK. Goddard Space Flight Center, Laboratory for High Energy Physics, Greenbelt, MD 20771, USA. 3 Department of Physics and Astronomy, University of Iowa, Iowa City, IA 52242, USA. 4 E.O. Hulbert Center for Space Research, Naval Research Laboratory, Code 4151, 4555 Overlook SW, Washington, DC 20375-5320, USA. 5 ISO Observatory, ESTEC, Postbus 299, 2200 AG Noordwijk, The Netherlands. 6 Astronomical Observatory, Jagiellonian University, Orla 171, 30-244 Cracow, Poland. 7 Department of Astronomy, Ohio State University, Columbus, OH 43210, USA. 8 N.Copernicus Astronomical Center, Bartycka 18, 00-716 Warsaw, Poland.
2 NASA
Abstract A series of ROSAT, ASCA and Compton Gamma-Ray Observatory (CGRO) observations of the Seyfert galaxy NGC 4151 were carried out during the period 1993 November 30 to December 13 as part of an intensive campaign to study the multiwavelength spectral characteristics of its short time-scale variability. In the softest X-ray bands monitored by ROSAT (0.1–0.4 keV, 0.5–1.0 keV) the source flux remained constant throughout the observing period. However, in an adjacent band (1.0–2.0 keV) significant variability was evident, the most obvious feature being a marked increase (factor 1.45) in the count rate over a timescale of ∼ 2 days commencing roughly 3 days into the monitoring period. In contrast, only a low amplitude of variability (< ∼ 10%) was measured in the four ASCA observations in the 2-10 keV band (but note that the first ASCA observation was performed somewhat after the onset of the flux increase seen by ROSAT). The count rates recorded by the Oriented Scintillation Spectrometer Experiment (OSSE) on CGRO are consistent with ±15% variations in the 50–150 keV gamma-ray band but there is no direct correspondence between the gamma-ray and soft X-ray light curves. The 0.1 to ∼ 300 keV spectrum of NGC 4151 is dominated by a hard power-law continuum which is cut-off at both high (∼ 90 keV) and low (∼ 4 keV) energy. A high energy cut-off is characteristic of a continuum generated by the process of thermal Comptonization whereas that at low energy arises from absorption in line-of-sight gas. In NGC 4151 this gas may be partially photoionized by the continuum source but still retains significant opacity below 1 keV. The observed soft X-ray variability may be the result of changes in the level of the underlying soft-hard X-ray continuum or changes in the line-of-sight absorption. The data marginally favour the former, in which case the difference between the soft X-ray and gammaray light curves implies a steepening of the continuum as the source brightens, consistent with earlier observations. As noted in earlier studies there is a soft excess below 1 keV which probably arises from more than one scattered and/or thermal components. The 1–2 keV soft X-ray and the ultraviolet continuum light curves (e.g. near 1440˚ A) show reasonably good correspondence, although the relative amplitude of the variations is much higher in the X-ray data. The observed ultraviolet to X-ray correlation has a similar slope to that established in earlier studies, although a significant residual ultraviolet flux is evident in the recent observations. A possible interpretation is that the X-ray to gamma-ray continuum is produced in a patchy dissipative corona above the surface of an accretion disk and that the correlated ultraviolet flux results from the reprocessing of part of this continuum by the disk. The residual ultraviolet flux may then arise from the reprocessing and/or the viscous heating of the disk.
Key words: Galaxies:individual:NGC 4151 – Galaxies:Seyfert – X-rays:galaxies – gamma-rays: observations.
1
Introduction
Over the last twenty years the Seyfert galaxy NGC 4151 has been extensively monitored in almost all the accessible wavebands. Its high energy spectrum has been particularly well studied for the obvious reason that the active galactic nucleus (AGN) at the centre of this nearby galaxy is one of the brightest extragalactic sources in the 2-100 keV band. Thus NGC 4151 has invariably been selected as a prime target for each new X-ray and gamma-ray mission. All this attention has provided a substantial observational database but unfortunately many of the properties of this AGN remain very unclear. Paradoxically this “archetypal” source has rather anomalous behaviour, for example NGC 4151 has a much harder continuum slope than is typical of many Seyfert 1/1.5 galaxies (but see Smith & Done 1995). The picture which has emerged is of a relatively low-luminosity AGN with a highly variable X-ray continuum, LX ∼ 2 − 20 × 1042 erg s−1 (2–10 keV). In the medium energy X-ray band there is evidence for a steepening of the continuum as the source brightens with an observed range of the energy spectral index of 0.3–0.7 (Perola et al. 1986; Fiore, Perola & Romano 1990; Yaqoob & Warwick 1991; Yaqoob et al. 1993). The 2–10 keV variability is rather sluggish compared to that exhibited by some Seyfert galaxies, with a typical flux doubling time-scale of ∼ 0.5 − 1 d. In addition there are occasional flaring events which occur on time-scales of days to weeks (Lawrence 1980; Perola et al. 1986; Yaqoob, Warwick & Pounds 1989; Yaqoob & Warwick 1991; Yaqoob et al. 1993). Recently the properties of NGC 4151 in the hard Xray/soft gamma-ray band have been better defined with the availability of qood quality data from the Oriented Scintillation Spectrometer Experiment (OSSE) on the Compton Gamma-Ray Observatory (CGRO). Zdziarski, Johnson & Magdziarz (1995) find the OSSE spectral data can be well described in terms of thermal Comptonization in a source having an intrinsic energy spectral index of 0.4–0.7 and a plasma temperature of ∼ 40−50 keV. Interestingly the 1991-1993 OSSE observations show the gamma-ray spectrum to have been almost constant in terms of the total flux and shape (Zdziarski, Johnson & Magdziarz 1995); the true nature of the wide-band spectral variability at present remains unclear. The hard X-ray continuum of NGC 4151 is cut-off at low energy (below ∼ 4 keV) due to photo-electric absorption in line-of-sight gas with hydrogen column density varying between ∼ 3–15 × 1022 cm−2 (Yaqoob & Warwick 1991; Yaqoob et al. 1993). An observation by the Solid-State Spectrometer abroad the Einstein observatory first showed the soft X-ray attenuation in the NGC 4151 spectrum to be incompatible with a uniform screen of cold solar abundance gas (Holt et al. 1980). The observed “soft excess” has subsequently been described in terms of both partial covering, a warm absorber and hidden spectral components (Holt et al. 1980; Yaqoob & Warwick 1991; Weaver et al. 1994a,b; Warwick, Done & Smith 1995). There is also a substantial flux observed below 1 keV which may arise, at least in part, from the spatial resolved soft X-ray source which coincides with the Extended Narrow-Line Region in NGC 4151 (Elvis, Briel & Henry 1983; Elvis et al. 1990; Morse et al. 1995). A final feature of the X-ray spectrum of NGC 4151 is an intense iron Kα emission line produced by the fluorescence of gas illuminated by the continuum X-rays. Recent ASCA spectra of NGC 4151 reveal the presence of both a narrow component and a redshifted broad component of this line with the latter possibly arising close to the central black hole in a putative accretion disk structure (see Yaqoob et al. 1995 for further details). Here we report the results of a series of observations carried out by ROSAT, ASCA and CGRO OSSE, which form part of an intensive campaign to study the short-timescale variability. Paper
I in this series (Crenshaw et al. 1995) reports the ultraviolet observations performed by the International Ultraviolet Explorer (IUE), which serve as the reference measurements for the whole programme. Paper II (Kaspi et al. 1995) similarly presents the results of the optical monitoring campaign. The current work (Paper III) concentrates on the high energy observations. Finally, Paper IV (Edelson et al. 1995) presents a detailed study of the short time-scale variability exhibited by NGC 4151 with emphasis on the multiwavelength characteristics of the phenomenon.
2
The Observations
A series of ROSAT, ASCA and CGRO OSSE observations of NGC 4151 were carried out during the period 1993 November 30 to December 13, as part of an intensive multiwavelength monitoring campaign. Details of the observing programs of the three satellites are summarised in Tables 1–3.
2.1
The ROSAT Observations
A total of 13 ROSAT observations were performed during the monitoring period, all with the Position Sensitive Proportional Counter (PSPC) in the focal plane of the X-ray telescope. A standard operational mode was employed in which the telescope’s pointing direction is “wobbled” during the observation (by ±3 arcmin on a time-scale of ∼ 400 seconds) in order to avoid the possibility of severe shadowing of an X-ray source by the coarse wire mesh which supports the PSPC window. The background count rate in the PSPC is relatively low and arises primarily due to cosmic-ray particles, scattered solar X-rays and the cosmic diffuse X-ray background (Snowden et al. 1992). For the present work we have applied a data rejection threshold of 170 count s−1 on the high energy particle detection rate monitored by the PSPC (the so-called Master Veto Rate) as a simple method of excluding time intervals when the particle background was unduly high. The resulting exposure times for the ROSAT observations are listed in Table 1. PSPC light curves and spectra for NGC 4151 and the BL Lac object 1E1207.9+3945 were extracted from the data using a source cell of radius 2.5 arcmin (the BL Lac object is located 5 arcmin to the north-west of NGC 4151 and serves as useful comparison source in the ROSAT observations). A corresponding background spectrum was taken from an annular region of radius 9–15 arcmin centred on NGC 4151 but from which discrete sources had been excised. After applying corrections for vignetting effects, the relative size of the background cell and the effective exposure, a background-subtracted PSPC spectrum was determined for each individual observation. Light curves were derived from the count rates in three spectral bands (corresponding to the energy ranges 0.1–0.4 keV, 0.5–1.0 keV and 1.0–2.0 keV), whereas for the spectral analysis the PSPC data were binned into 27 spectral channels covering the full 0.1–2.0 keV energy range (this represents a significant over-sampling of the spectra in comparison to the energy resolution of the PSPC). The ROSAT PSPC spectra for NGC 4151 were analysed using a standard model fitting procedure (xspec version 8.50) and the DRM31 instrument response matrix (which is currently recommended for observations performed in the ROSAT AO-3 period). A 1% systematic error was added to the data to take account of the uncertainties in the response matrix.
2.2
The ASCA Observations
Four observations of NGC 4151 were carried out with the two Solid-State Imaging Spectrometers (SIS-0 and SIS-1) and two Gas Imaging Spectrometers (GIS-2 and GIS-3) on ASCA (Tanaka, Inoue & Holt 1994) as part of the monitoring campaign (see Table 2). The SIS data, which were taken in 2-CCD FAINT and BRIGHT modes, were first all converted to BRIGHT mode and then events of grade 0, 2, 3 and 4 were used in the subsequent analysis. Events were accumulated from circular regions centred on NGC 4151 with extraction radii of 3.9′ and 3.4′ for SIS-0 and SIS-1 respectively, and 6.0′ for GIS-2 and GIS-3. Event lists were also obtained for the off-source regions. For the SIS, this included all events from the source chip farther than 6.0′ from NGC 4151 and, for the GIS, from an annulus of radius 9′ –13′ centred on NGC 4151 (events from the BL Lac were excluded). Both the source and background lists were examined for anomalous events which were then rejected; this is a very effective way of removing large numbers of “flickering” pixels and light-leakage events. Events recorded during spacecraft passages through the South Atlantic Anomaly, during periods when the magnetic cut-off rigidity was less than 7.5 GeV/c2 and at times when the Earth elevation angle was less than 5◦ were also rejected. Finally, any remaining “hot” pixels were removed by examining the count/pixel distributions in a large number of annular rings. X-ray spectra were made from the resulting cleaned source and background event lists for each detector corresponding to the four observing periods. Summed spectra corresponding to the entire observing period were also produced. As a check the analysis was repeated with more severe selection criteria but this resulted in no significant improvement in the quality of spectra obtained. For the purpose of the present paper, we have analysed only the SIS-0 data. These are the “best” data in the sense that SIS-0 is better calibrated and the source appears closer to the XRT optical axis, giving higher count rates. There is, in fact, very good agreement amongst the four ASCA instruments for this set of observations. Yaqoob et al. (1995) also show that the SIS data require no correction to the energy scale. The present observations were made shortly after the end of the Performance Verification Phase at which point the degradation of the CCD charge transfer efficiency had no measurable impact on the data quality. Fluctuations in the “zero” level of the energy scale by 10–30 eV (Dark Frame Error) cause a degradation of the energy resolution rather than a measurable non-zero offset.
2.3
The CGRO OSSE Observations
CGRO OSSE monitored NGC 4151 continuously over a period of 13 days during the campaign (see Table 3). OSSE observations consist of a sequence of two-minute integrations on the source field alternated with similar offset-pointed background measurements. Background estimation is achieved by performing a quadratic fit to three or four contiguous two-minute background accumulations of satisfactory data quality taken from the immediate temporal vicinity of each two-minute on-source accumulation. The resulting two-minute background-subtracted spectra are then summed over the period of a day. In the present analysis the data processing and screening produced a total on-source livetime of 4.9×105 detector-seconds. A similar duration of offset-pointed background observations were used in creating the background-subtracted spectra. (See Johnson et al. 1993 for a more detailed description of OSSE performance and analysis.) The source was detected at a significance level of greater than 12σ per day in the energy range 50−150 keV. The daily average flux in this band is seen to vary by ±15% during the observation. The average spectrum of NGC 4151 for the whole observing period was created by summing the daily spectra for all four detectors. For analysis by xspec the OSSE data were binned into 14 spectral channels covering the energy range 50 keV to 1 MeV. Systematic errors were
added to the statistical errors before the spectral fitting. These systematics were estimated from uncertainties in the low energy calibration and in the detector response and are significant below 100 keV with a peak of ∼ 10% of the statistical error at 50 keV.
3
The ROSAT results
Figure 1 shows the PSPC count rates measured in 3 spectral bands for both NGC 4151 and the nearby BL Lac object, 1E1207.9+3945, in each of the 13 ROSAT observations. In the case of 1E1207.9+3945 there is no clear evidence for variability over the 10-day monitoring period. Thus, a χ2 test based on the hypothesis that the BL Lac source flux is constant gave χ2 values of 7.0, 12.3 and 18.1 for 12 degrees-of-freedom (d-o-f) for the soft, medium and hard ROSAT bands respectively. A similar situation pertains for the soft and medium band light curves of NGC 4151 (the constant source hypothesis gave χ2 = 6.8 and 6.6 with 12 d-o-f for the two bands respectively). However, in the 1.0–2.0 keV ROSAT band variability is evident (χ2 = 215 for 12 d-o-f) in the form of a marked change in the count rate commencing roughly 3 days into the monitoring campaign at which point the signal increases by about a factor 2 (minimum to maximum) over a period of two days. Although, not strictly consistent with a constant count rate, the first six ROSAT observations (hereafter group A) have a mean count rate in the hard band of 0.22 count s−1 , whereas the last five observations (group B) have a mean count rate of 0.32 count s−1 ; this corresponds to a 45% stepwise flux increase. Our preliminary analysis of the spectral variability exhibited by NGC 4151 is based on the combined datasets, defined as group A and B above. The disparate light curves of NGC 4151 in the soft/medium and hard ROSAT bands (Figure 1) provide clear evidence for a spectral change. This point is further illustrated in Figure 2, which shows the ratio of pulse height spectra for the group A and B datasets; the variability is clearly restricted to spectral channels above ∼ 1 keV. In order to investigate the nature of this soft X-ray variability we have employed a simple phenomenological model with two emission components, namely a thermal bremsstrahlung continuum representative of the bulk of the soft X-ray flux and a hard power-law continuum which dominates above ∼ 1 keV. The former component is presumed to suffer only light absorption due to a modest line-of-sight column density, NHgal , which arises in our own galaxy, whereas the latter is relatively heavily absorbed by a gas column, NH 1 , intrinsic to NGC 4151 (and presumably located in its active nucleus). In NGC 4151 there is evidence for a spectral index–flux correlation in the sense that the Xray spectrum softens as the source brightens. The energy index, α, varies between 0.3 and 0.7 at which point the correlation appears to “saturate” (Yaqoob & Warwick 1991; Yaqoob et al. 1993). As we do not know the energy index a priori (the ROSAT PSPC has insufficient spectral resolution and bandwidth to constrain this parameter usefully), in the present analysis we have fixed the value of α at 0.5. The spectral modelling procedure we have adopted is to fit simultaneously the PSPC spectra from the group A and B datasets with the temperature, kTbrem , normalisation, Abrem , and absorption, NHgal , of the thermal bremsstrahlung component constrained to the same values for the two datasets. In contrast, the two free parameters defining the power-law continuum (normalisation, A, and absorption, NH ) were not tied across the two datasets, thus catering for the variability evident above 1 keV. The results from the simultaneous 1 Throughout this paper we have used the element abundances and absorption cross-sections adopted by Morrison & McCammon (1983)
fit to the group A and B datasets (involving a total of 7 free parameters) are given in Table 4. (The errors quoted in Table 4 and elsewhere in this paper represent 90 percent confidence limits for one interesting parameter, see Lampton, Margon & Bowyer, 1976). From these results it may be concluded that the gross variability in the ROSAT band is attributable mainly to a change in the normalisation of the hard power-law continuum (by a factor of ∼ 2.4) rather than to a variation in the absorption parameter, NH . When the value of NH was tied across the two datasets, the minimum χ2 for the joint fits showed little variation compared to that quoted in Table 4. To further illustrate the influence of variations in the level of the hard continuum and/or column density changes, we have determined the variation in the minimum χ2 for models in which the values of A and NH were constrained to fixed ratios across the group A and group B datasets. The result, shown in Figure 3, emphasises the point that although column density variations cannot be excluded (of magnitude < ∼ 20%) a significant change in the normalisation of the hard power-law is definitely required by the data. Thus, here we take the view that the variability revealed by the ROSAT observations is largely attributable to variations in the level of the hard X-ray continuum in NGC 4151. The two component spectral model discussed above was also applied to the spectra from each of the individual ROSAT observations but with only the normalisation of the hard power-law considered as a free parameter (the other parameters of the model being fixed at their global best-fit values). These spectral fits were then used to derive the corresponding X-ray fluxes in the 1–2 keV band. The results are tabulated in Table 1 and also plotted as a light curve in Figure 4. Note that in the model discussed above the contribution to the flux in the 1–2 keV band from the (non-variable) thermal bremsstrahlung component is 9.2 × 10−13 erg cm−2 s−1 . A further point concerns the best fitting parameters for the thermal bremsstrahlung component. The values quoted in Table 4 are in fair agreement (within the errors) with those obtained in June 1991 in an earlier ROSAT PSPC observation (Warwick et al. 1995); however, the observed 0.2–1.0 keV flux of ∼ 4.9 × 10−12 erg cm−2 s−1 is approximately 30% higher than that measured previously. We note that recent ASCA observations (Weaver et al. 1994b) have provided evidence for variability (over a time-scale of ∼ 6 months) in the the soft X-ray spectrum of NGC 4151 below 1 keV. Clearly such variability gives an important clue as to the origin of the soft X-ray flux in NGC 4151 and can be used to set limits on what fraction of the emission originates in extended thermal or electron scattered components. However, at present we are cautious in ascribing the apparent variations in the sub-1 keV PSPC flux (over a time-scale of 2.5 years) to intrinsic source variability in NGC 4151 since PSPC calibration uncertainties (particularly in the 0.1–0.4 keV band) may account, at least in part, for the differences in the flux measurements. Finally we note that in the above analysis the selection of a different energy index other than α = 0.5 but in the range 0.3–0.7 has only a marginal effect on the spectral fitting results. The conclusions are essentially unaltered for any value of α within this specified range.
4
The ASCA Results
Comparison of the count rates measured on NGC 4151 from the SIS-0 telescope during the four ASCA observations provides evidence for low amplitude (< ∼ 10%) but significant variability above 1 keV. Unfortunately the first ASCA observation was performed on 1993 December 4, somewhat after the onset of the largest amplitude change seen by ROSAT.
Our preliminary analysis of the ASCA SIS-0 spectra for the energy band 0.4–10 keV is based on a partial covering description of the complex absorption feature which dominates the 1–5 keV spectrum of NGC 4151. Specifically we use the “dual-absorber” discussed by Weaver et al. (1994b) in which a fraction CF of the hard X-ray flux suffers absorption in a column density, NH1 , with the remaining 1 − CF fraction subject to a somewhat lower (but still substantial) column density, NH2 . Further parameters are then required to represent the iron-K line and any additional iron edge features2 apparent in the energy range 5–8 keV. For simplicity the present analysis only includes a narrow iron Kα line with an intrinsic width σKα = 0.07 keV (see Yaqoob et al 1995 for a more detailed discussion of the iron features in ASCA spectra of NGC 4151). In addition the bulk of the soft X-ray flux below 1 keV was initially modelled as a thermal bremsstrahlung continuum absorbed by a column density NHgal (c.f. section 3). At this point NHgal was fixed at 2.1 × 1020 cm−2 which is the Galactic HI column density measured in the direction of NGC 4151 (and is also reasonably consistent with the value for NHgal obtained in the earlier ROSAT analysis). As in the previous section we adopt the value α = 0.5 for the energy index of the hard continuum. Initially we considered a simultaneous fit to the four ASCA spectra with only the normalisation of the hard power-law continuum, A, free to vary across the four datasets; details of the bestfitting model are given in Table 5. This analysis shows that the level of variability in the 2 underlying X-ray continuum, as measured by its normalisation, was < ∼ 10%. The minimum χ of 1371 (1052 d-o-f) demonstrates that this simplified partial covering model does not provide a full description of the ASCA spectra. There is evidence for additional spectral features and subtle spectral variability across the four ASCA datasets. This is illustrated in Figure 5 which shows the ratio of the four ASCA spectra to the best-fitting spectral form (i.e. the model defined in Table 5). Residual features are apparent in all the spectra; for example, three out of the four ASCA spectra show evidence for a narrow line feature at ∼ 0.89 keV with an average flux of 0.8 × 10−4 photon s−1 cm−2 and an equivalent width of ∼ 30 eV. (A similar line feature was noted in previous ASCA observations by Weaver et al. 1994b). However, across the full 0.5–10 keV band there is no clear evidence for systematic spectral variations. It is possible that some spectral variability results from changes in the complex absorber. To investigate this we have carried out a spectral fitting analysis with the parameters NH1 , CF and NH2 , as well as the continuum normalisation, free to vary across the individual observations. The result was a significant reduction in the minimum χ2 by 69 to 1302, with 1043 d-o-f; however, the derived error ranges on the individual parameters were large and in all cases consistent with the average parameter values obtained earlier (Table 5). We note that subtle changes in the form of the absorption on a timescale of days have been reported in previous studies of NGC 4151 (e.g. Weaver et al. 1994a; Warwick et al. 1995). The derived fluxes from the ASCA observations in the 1–2 and 2–10 keV energy bands are listed in Table 2 and also plotted in Figure 4. An important point to note is that there is rather good agreement between the ROSAT and ASCA measurements of the 1–2 keV flux for those observations which overlap in time. Significant variability is apparent in the combined ROSAT–ASCA light curve. Unfortunately the coverage of the 2–10 keV band in the ASCA (only) measurements is too sparse to confirm the large amplitude variability as a broad-band effect. 2
Iron-K edge features arising from a solar abundance concentration of iron atoms in the gas columns NH1 and NH2 are automatically included in the spectral fitting.
5
The CGRO OSSE results
The light curve in the 50–150 keV band derived from the CGRO OSSE observations is shown in Figure 6. These data are inconsistent with a constant count rate (χ2 = 28.4 for 11 d-o-f.) with variations from the mean of up to ±15% peak-to-peak (∼ 10% rms). Similar variations have been reported by Maisack et al. (1993) from a previous OSSE observation of this source. The present gamma-ray measurements fail to show any significant variation corresponding to the flux increase apparent in the 1–2 keV ROSAT band. Interpreting the latter as a factor ∼ 2.4 increase in the power-law continuum at ∼ 1 keV (section 3) and the former as a continuum increase at ∼ 100 keV of no more than 15% (estimated for the relevant periods from Figure 6) implies a change in the spectral index of the underlying hard power-law continuum ∆α ≈ 0.15. We note that this is compatible with the spectral softening as the 2–10 keV flux increases which has previously been reported for NGC 4151 (Perola et al. 1986; Fiore, Perola & Romano 1990; Yaqoob & Warwick 1991; Yaqoob et al. 1993). There is no evidence for spectral variability within the set of OSSE observations, thus we have considered only the mean spectrum. This is not well described by a single power-law continuum (spectral fitting of a two parameter model gave a minimum χ2 = 42.6 for 12 d-o-f.). With this model the spectral index, α = 1.61 ± 0.08, is much softer than that observed at lower energies, indicating that the spectrum must break at an intermediate energy. Thus, we have adopted a spectral model in which a hard power-law continuum is modified by an exponential cut-off of the form exp(−E/Ec ), where Ec is the characteristic energy. If we fix the hard power-law 2 slope at α = 0.5 (see Section 3 and 4) then Ec = 87+8 −7 keV and χ = 7.9 (12 d-o-f.). In this case the best-fit normalisation (at 1 keV) is 7.1 ± 0.7 × 10−2 photon cm−2 s−1 keV−1 , in fairly close agreement with the values derived from the fits to the ASCA data (see Table 5). Also, the observed 50–150 keV flux values for each of the individual OSSE observations were calculated on the basis of this spectral model; these are listed in Table 3. Alternatively the OSSE data can be fitted with the thermal Comptonization model of Titarchuk & Mastichiadis (1994). We obtained best fit parameter values of kT = 50+25 −12 keV and α = +0.21 0.65−0.18 , where α is the asymptotic low-energy power-law index (we use this parameter rather than the geometry-dependent Thomson optical depth). In this case the χ2 was 7.0 for 11 d-o-f. Very similar results have been obtained by Zdziarski, Johnson & Magdziarz (1995) who provide a detailed critique of the Titarchuk & Mastichiadis model and also survey the available set of OSSE and other hard X-ray/ gamma-ray observations of NGC 4151.
6
The wide band X-ray to gamma-ray spectrum of NGC 4151
Spectral coverage from 0.1 keV to ∼ 300 keV is obtained from the combined ROSAT, ASCA and CGRO OSSE data sets. Here we consider only the first ASCA observation together with the mean spectrum from the near simultaneous ROSAT observations (observations 8 and 9). However, we use the full set of CGRO OSSE observations since there is no convincing evidence for spectral variability in these data. The quasi-simultaneous data were initially fitted with the partial covering model described in Section 4 with the addition of the exponential cut-off to the hard power-law continuum (Section 5). The spectral index was again fixed at α = 0.5. Table 6 gives the details of the best-fitting
parameters (the iron Kα line parameters are not quoted since the values are little changed from Table 5). This fit to the combined data demonstrates that there is reasonable agreement between the three instruments as illustrated in Figure 7a, which shows the measured count rate spectra compared with the best-fit model. The excellent correspondence between the ROSAT PSPC and ASCA SIS-0 data in the 1–2 keV range suggests that the relative cross-calibration of the two instruments, albeit in this limited energy range, is good to better than 10% (thus validating the combined ROSAT/ASCA 1–2 keV light curve in Figure 4). However, below 1 keV, there is evidence from the spectral fitting residuals for systematic differences between the ROSAT and ASCA datasets, which presumably relate to calibration uncertainties (see Figure 7b). The derived input spectrum corresponding to the above partial covering model is shown in Figure 7c. This figure illustrates the main characteristic of the soft X-ray to gamma-ray spectrum of NGC 4151, namely its domination by a hard power-law continuum which is cut-off at both high (∼ 90 keV) and low (∼ 4 keV) energy. The high energy cut-off is characteristic of a continuum generated by the process of thermal Comptonization (see section 5; Titarchuk & Mastichiadis 1994; Zdziarski, Johnson & Magdziarz 1995) whereas that at low energy arises from absorption in a substantial column density (∼ 1023 cm−2 ) of relatively cool gas in the line-of-sight to the continuum X-ray source. As noted in many earlier studies (and in previous sections) there is also a soft excess flux below 1 keV which is probably due to the superposition of a number of seperate spectral components (although so far we have used only a single temperature thermal bremsstrahlung representation). In the remainder of this section we comment on various aspects of the high energy spectrum of NGC 4151.
6.1
The soft excess flux below 1 keV
The thermal bremsstrahlung description of the soft X-ray spectrum of NGC 4151 is not satisfactory since line emission dominates the cooling of thermal plasmas in the temperature range under consideration. In addition the steep slope of the 0.1–1.0 keV spectrum of NGC 4151 rules out an origin for this flux solely as electron scattering of the hard X-ray continuum. To illustrate this last point we have repeated the above spectral fitting with the thermal bremsstrahlung component replaced by a soft power-law component (with energy index, αsof t and normalisation Asof t ). The results detailed in Table 6 show that αsof t ≈ 1.3, which is a much steeper slope than is measured for the underlying X-ray continuum (albeit at energies well above 1 keV). A plausible physical description of the 0.1–1.0 keV spectrum of NGC 4151 requires at least two spectral components. Comparison of Tables 4 and 5 shows that the ROSAT data alone require a lower temperature for the thermal bremsstrahlung component than when fitting the ASCA data alone (i.e. ∼ 0.4 keV as opposed to ∼ 1.3 keV). This suggests that an additional “ultra-soft” component is required by the data as previously argued by Weaver et al. (1994b) and Warwick et al. (1995). The ASCA data are not compatible with a sizeable fraction of the 0.5–1.0 keV soft X-ray flux originating as thermal emission in a solar abundance gas (this is based on the lack of prominent O VII, O VIII and Fe-L emission lines). However, it is plausible that the bulk of the flux observed in this narrow spectral range is predominantly the extrapolated hard X-ray continuum seen as a result of pure electron scattering in a medium extending along the symmetry axis of the source (by analogy with the situation pertaining in NGC 1068, see Marshall et al. 1993 and references therein). Thus, a possible two component model involves a power-law continuum (with spectral index similar to that of the hard power-law continuum) plus an ultra-soft component. The latter
may be modelled as either a ∼ 106 K thermal plasma (perhaps associated with the extended source close to the nucleus of NGC 4151; Elvis et al. 1983; Morse et al. 1995) or a scattered black-body component (possibly originating in the hot inner regions of a putative accretion disc). For a more detailed discussion see Weaver et al. (1994a,b) and Warwick et al. (1995).
6.2
Complex absorption in the 1–5 keV band
Although partial covering provides a viable description of the complex absorption observed in the 1–5 keV spectrum of NGC 4151, an alternative and perhaps more physically realistic approach is to utilise the reduced soft X-ray opacity exhibited by a partially ionized medium. This “warm absorber” model has previously been considered in the NGC 4151 context by a number of authors (Yaqoob & Warwick 1991; Netzer 1993; Weaver et al. 1994a,b; Warwick et al. 1995). In modelling the composite ROSAT–ASCA–OSSE spectrum with a warm absorber we replace the three partial covering parameters with just two, namely the column density of the medium NH and the ionization parameter ξ (where ξ = Lion /nr 2 and the ionizing luminosity, Lion , is obtained by integrating the extrapolated continuum between 5 eV and 300 keV; see Warwick et al. 1995 and Zdziarski et al. 1995). We initially revert to the case where the only additional soft component is the scattered powerlaw mentioned above (with the spectral index set equal to that of the hard X-ray continuum). With this model a reasonable fit was obtained (χ2 = 412/ 302 d-o-f) with NH = 9.6×1022 cm−2 and ξ = 28; the corresponding input model spectrum is shown in Figure 8a. However, there is an immediate difficulty with this particular description. ¿From the ROSAT observations it is clear that there was no variability below ∼ 1 keV at the time the hard X-ray continuum varied substantially. This essentially rules out the situation (as in Figure 8a) in which a significant fraction of the 0.1–1.0 keV flux arises due to the leakage of the underlying X-ray continuum through the warm absorbing medium (in which there is reduced soft X-ray opacity as a result of photoionization, e.g. Krolik & Kallman 1987), since in such a model the ultra-soft flux must track the continuum variation. This problem can be circumvented by invoking a somewhat lower value for the ionization parameter or, alternatively, including an additional (cold) column density to suppress the transmitted soft X-ray flux. Either way, at least one additional soft spectral component is required to account for the ultra-soft excess flux. By way of illustration, Figure 8b shows the input spectrum for a revised warm absorber model (which gave χ2 = 396/300 d-o-f) where NH = 8.2 × 1022 cm−2 and ξ = 14 and which includes an additional ∼ 106 K Raymond Smith thermal component (with solar element abundances assumed). The decrease in the ionization parameter by about a factor of two, greatly reduces the leakage of the hard continuum below 1 keV with the deficit being made up, of course, by the thermal emission. Thus in this setting the scattered power-law and thermal component account for the bulk of the soft X-ray flux below ∼ 1 keV, hence explaining the lack of variability in the 0.1–1.0 keV band. It should be noted that there are many unresolved questions relating to both the 0.1–1 keV spectrum and the complex absorption apparent in NGC 4151. For example, the lack of a prominent O VII line at 570 eV, O VIII Lα line at 650 eV and the Fe-L blend near 1 keV in the ASCA spectra limits the contribution of thermal sources with temperature in the range 106 −107 K. This is already a problem for the model in Figure 8b where the observed equivalent width of the OVII line is ∼ 150 eV whereas the ASCA spectrum gives an upper limit close to 50 eV for an isolated narrow line feature at an energy of 570 eV. This is unfortunate since the 106 K thermal component can account for roughly ∼ 25% of the count rate measured by the ROSAT
HRI on NGC 4151, which is comparable to the fraction attributed to the extended emission (Morse et al. 1995). Since the extended emission is very likely to be thermal in origin (unless the X-ray emission is strongly anisotropic - see Morse et al.) the lack of strong line emission below 1 keV could indicate either gas with a low metal abundance or gas which is over-ionized (as a result of the incident photoionizing flux) relative to coronal equilibrium conditions. This problem is even more acute if the temperature of the extended emission is closer to 107 K.
6.3
The hard X-ray/gamma-ray continuum in NGC 4151
The earlier spectral analysis has been largely based on an assumed value for the energy index of the hard X-ray continuum, α = 0.5. Compatibility with the combined ASCA SIS-0 and CGRO OSSE spectral data is then obtained with the inclusion of an exponential cut-off term with a characteristic energy Ec ≈ 90 keV. If the spectral index is allowed to be a free parameter in the original partial covering model discussed above, the resulting best fit gives α ≈ 0.45, in line with our assumption. ¿From the results in Table 6 we obtain an absorption-corrected flux in the 2–10 keV band of 3.6 × 10−10 erg cm−2 s−1 . This corresponds to a particularly bright state of the source with reference to the compilation of results from EXOSAT and Ginga reported in Yaqoob et al. (1993). Figure 2 of Yaqoob et al. (1993) shows that such X-ray bright states are more typically characterised by α = 0.6 − 0.7. However, there is no real inconsistency with the present measurements since the continuum slope of NGC 4151 is not particularly well constrained by the combination of the ASCA and OSSE spectra (mainly because the strong absorption significantly reduces the effective bandwidth for the former). As a minor point we also note that exponential form for the high energy cut-off serves to steepen the effective spectral slope in the medium energy band by ∆α ∼ 0.05. The thermal Comptonization model discussed in section 5 and, in the NGC 4151 context by Maisack et al. (1993), Titarchuk & Mastichiadis (1994) and Zdziarski, Johnson & Magdziarz (1995), provides a physical basis for the observed continuum form in this AGN. Specifically the model of Titarchuk & Mastichiadis (1994) when fitted to the composite data (again in the +0.03 context of the partial covering model) gives kT = 49+8 −6 keV and α = 0.64−0.02 , values which are very similar to those obtained from the OSSE only spectral fits. Here we derive a somewhat steeper value for the asymptotic low-energy spectral index than was obtained with the cut-off power-law model due presumably to differences in the exact form of the two spectral models within the limited spectral coverage. As emphasised by Zdziarski, Johnson & Magdziarz (1995), the hard spectrum of NGC 4151 implies that the hot Comptonized plasma is starved of soft photons, a situation which could arise, for example, in very patchy corona situated above the surface of an accretion disk. The gap in spectral coverage between the ASCA and OSSE data prevents a very sensitive test for the presence of a Compton reflection component (e.g. Zdziarski, Johnson & Magdziarz 1995). However, if we fix α = 0.9 (a typical value for Seyfert 1 galaxies - see Nandra & Pounds 1994) in the cut-off power-law model and include a reflection component (with cos i set to 0.45), then the best fit to the observed spectrum requires a relative normalisation for the reflection, R = 2.2 (see Magdziarz & Zdziarski 1995). Since Ginga data for this object strongly rule out such a strong reflection component (Maisack & Yaqoob 1991; Yaqoob et al. 1993), this in turn implies that the continuum spectrum of NGC 4151 is indeed atypical of Seyfert 1 galaxies as a class. However, an interesting point is that R = 2.2 implies an equivalent width of the iron K-line of ∼ 300 eV (for a solar abundance reflector) which is comparable to the equivalent width
observed when a full account is taken of both broad and narrow line components in the 5–8 keV band (see Yaqoob et al. 1995 for a full discussion of the iron-line properties of NGC 4151).
7
The short-timescale X-ray variability in NGC 4151
The primary objective of the program discussed here (and in Papers I, II and IV) is the study the short time-scale spectral variability exhibited by NGC 4151. The derived light curves in the 1–2 keV, 2–10 keV and 50–150 keV bands are shown in Figures 4 and 6. Also Table 7 summarises the basic properties of the variability in each of these bands (we exclude the 0.1-1.0 keV range since there is no evidence for variability in this band). For each energy range the table lists the mean flux, the fractional measurement error (which is the average value of the flux error divided by the mean flux) and the fractional variability (which is the standard deviation of the fluxes divided by the mean flux). As noted earlier, the largest amplitude variations were recorded in the 1–2 keV band; however, an important qualification is that the ASCA 2–10 keV monitoring did not commence until after the onset of the significant flux increase observed by ROSAT in the 1–2 keV band. Although the soft X-ray variability may, in principle, arise from either changes in the level of the underlying continuum or changes in the line-of-sight absorption, the ROSAT data marginally favour the former possibility (see section 3), and we focus on this interpretation in this paper. The difference between the soft X-ray and gamma-ray light curves requires a steepening of the continuum as the source brightens in X-rays, and suggests that the wide-band variability takes the form of a “pivoting” of the continuum around a point near ∼ 100 keV (see section 5; Yaqoob & Warwick 1991; Zdziarski, Johnson & Magdziarz 1995). ˚ light curves for NGC 4151 Figure 9 compares the soft X-ray 1–2 keV and the ultraviolet 1440 A (see Paper I for details of the IUE measurements). A detailed investigation of the temporal relation between the soft X-ray and longer wavelength variability is presented in Paper IV (Edelson et al. 1995) and here we restricted our attention to the form of the ultraviolet to X-ray correlation which is apparent in Figure 9. The main feature of the soft X-ray light curve, namely the marked flux increase which commenced roughly 3 days into the observation, coincides with a rather similar variation in the ultraviolet continuum band. However, the relative amplitude of the X-ray variability is significantly higher than is seen even in the shortest wavelength ultraviolet bands (compare the results in Table 7 with those given in Table 5 of Paper I). It also apparent that the correlation between the ultraviolet and soft X-rays is far from perfect; for example, the ultraviolet flux at 1440 ˚ A shows a gradual decline over the second half of the observing period which was not mirrored in soft X-ray band. In order to investigate the form of the correlation between the X-ray and the ultraviolet continuum flux we have extracted the 1440 ˚ A fluxes averaged over a period of ±0.3 d corresponding to each ROSAT measurement. In addition the observed 1–2 keV fluxes were scaled to absorption-corrected 2–10 keV continuum fluxes, F c (2-10 keV), based on the spectral model in Table 6 (thermal bremsstrahlung case). In this process the contribution of the (non-varying) thermal bremsstrahlung component is automatically removed. Figure 10 shows the resulting correlation plot. If we exclude the last ROSAT observation (which was taken 3 days after the main group of observations and at the time when the decline in the ultraviolet flux had no clear correspondence in the X-ray data), then there is good correlation between the two wavebands (correlation coefficient = 0.86). The X-ray/ultraviolet correlation seen in the current observations may be compared to that found earlier by Perola et al. (1986) in a comparison of EXOSAT and IUE measurements (represented by the dotted line in Figure 10). Perola & Piro (1994) have discussed the earlier
data in terms of a model in which a central spherical source illuminates a thin accretion disk with a hard continuum which is reprocessed to give rise to the correlated ultraviolet flux. In this model the heat input due to the incident radiation is presumed to dominate that due to viscous heating of the disk. This is consistent with the fact that the X-ray/ultraviolet correlation observed by Perola et al. (1986) implies almost no residual (uncorrelated) ultraviolet flux. Perola & Piro (1994) found that in order to match the observed ultraviolet flux and ultraviolet spectral index it was necessary to assume that the gamma-ray continuum extended out to ∼ 4 MeV, contrary to current CGRO OSSE results (e.g. section 6; Zdziarski, Johnson & Magdziarz, 1995). However, in a recent paper Zdziarski & Magdziarz (1995) argue that if the source of the X-ray to gamma-ray continuum is a patchy dissipative corona situated above the surface of the accretion disk, then the problem of powering the observed ultraviolet flux is largely overcome. Although the slope of the X-ray/ultraviolet correlation measured in the present observations (Figure 10) is similar to that measured by Perola et al. (1986), a very considerable residual ultraviolet flux is now predicted when the 2–10 keV X-ray emission is extrapolated to zero. It is possible that this additional ultraviolet flux is evidence of an enhancement in the viscous heating processes in the disk. However, in the reprocessing scenario it should be noted that although the soft X-ray continuum is highly variable, at gamma-ray energies, where the bulk of the energy resides, the continuum is much more stable. Thus an alternative possibility is that the residual ultraviolet component arises from the reprocessing of the high energy end of the continuum. This would suggest that the different correlations evident in Figure 10 mirror a significant change in the form of the gamma-ray spectrum of NGC 4151 between 1986 and 1993.
8
Conclusions
We have presented a new analysis of the high energy properties of NGC 4151 based on ROSAT, ASCA and CGRO OSSE data. The soft X-ray to gamma-ray spectrum of NGC 4151 measured in simultaneous observations is dominated by a hard power-law continuum, which is cut-off at both high (∼ 90 keV) and low (∼ 4 keV) energy. The high energy cut-off may originate from the process of thermal Comptonization whereas that at low energy certainly arises from absorption in line-of-sight gas. In NGC 4151 this gas may be partially photoionized by the continuum source but still retains significant opacity below 1 keV. As noted in earlier studies there is a soft excess below 1 keV which probably arises from more than one scattered and/or thermal components. Of the various X-ray and gamma-ray wavebands monitored in the present campaign the largest amplitude of variability (∼ 30% rms) is observed in the 1.0–2.0 keV range. The most obvious feature in the soft X-ray light curve is a marked increase (factor 1.45) in the count rate over a timescale of ∼ 2 days. During the same period the count rates recorded by OSSE in the 50–150 keV gamma-ray band show no more than ±15% changes and no clear correspondence with the soft X-ray variations. The difference between the soft X-ray and gamma-ray light curves may be interpreted as a steepening of the continuum as the source brightens in X-rays and suggests the wide-band variability takes the form of a “pivoting” of the continuum around a point near ∼ 100 keV. In contrast, the 1–2 keV light curve does show a reasonable correspondence with that observed in the ultraviolet continuum, although the relative amplitude of the variations is again much higher in the X-ray data. A possible interpretation of the correlated ultraviolet to X-ray variability is that the hard (X-ray to gamma-ray) continuum in NGC 4151 is produced in a patchy
dissipative corona above the surface of an accretion disk and that the ultraviolet flux results from the reprocessing of part of this continuum by the disk. The origin of the uncorrelated (residual) ultraviolet flux is uncertain but reprocessing and/or the viscous heating of the disk are possibilities.
9
Acknowledgements
This research has been supported in part by the NASA grants NAG5-2439, NAG5-1813, NAGW3129 and NAS5-30960. David Smith acknowledges support from PPARC in the form of a research studentship.
10
References
Crenshaw, D.M. et al., 1995, ApJ, submitted (Paper I) Edelson, R.A. et al., 1995, ApJ, submitted (Paper IV) Elvis, M., Briel, U.G., Henry, J.P., 1983, ApJ, 268, 105 Elvis, M., Fassnacht, C., Wilson, A.S., Briel, U., 1990, ApJ, 361, 459 Fiore, F., Perola, G.C., Romano, M., 1990, MNRAS, 243, 522 Holt, S.S., Mushotzky, R.F., Becker, R.H., Boldt, E.A., Serlemitsos, P.J., Szymkowiak, A.E., White, N.E., 1980, ApJ, 241, L13 Johnson, W.N., Kinzer, R.L., Kurfess, J.D., Strickman, M.S., Purcell, W.R., Grabelsky, D.A., Ulmer, M.P., Hillis, D.A., Jung, G.V., Cameron, R.A., 1993, ApJS, 86, 693 Kaspi, S., et al., 1995, ApJ, submitted (Paper II) Krolik, J.H., Kallman, T.R., 1987, ApJ, 320, L5 Lampton, M., Margon, B., Bowyer, S., 1976, ApJ, 208, 177 Lawrence, A., 1980, MNRAS, 192, 83 Magdziarz, P., Zdziarski, A. A., 1995, MNRAS, 273, 837 Maisack, M., Yaqoob, T., 1991, AA, 249, 25 Maisack, M. et al., 1993, ApJ, 407, L61 Marshall, F.E. et al., 1993, ApJ, 405, 168 Morse, J.A., Wilson, A.S., Elvis, M., Weaver, K.A., 1995, ApJ, 439, 121 Morrison, R. McCammon, D., 1983, ApJ, 270, 119 Nandra, K., Pounds, K., 1994, MNRAS, 268, 405 Netzer, H., 1993, ApJ, 411, 594 Perola, G.C. et al., 1986, ApJ, 306, 508 Perola, G.C, Piro, L., 1994, AA, 281, 7 Smith, D.A., Done, C., 1995, MNRAS, in press Snowden, S.L., Plucinsky, P.P., Briel, U., Hasinger, G., Pfeffermann, E., 1992, ApJ, 393, 819 Tanaka, Y., Inoue, H., Holt, S.S., 1994, PASJ, 46, L37 Titarchuk, L., Mastichiadis, A., 1994, ApJ, 433, L33 Warwick, R.S., Done, C., Smith, D.A., 1995, MNRAS, 275, 1003 Weaver, K.A., Mushotzky, R.F., Arnaud, K.A, Serlemitsos, P.J, Marshall, F.E., Petre, R., Jahoda, K.M., Smale, A.P., 1994a, ApJ, 423, 621 Weaver, K.A., Yaqoob, T., Holt, S.S., Mushotzky, R.F., Matsuoka, M., Yamuchi, M., 1994b, ApJ, 436, L27 Yaqoob, T., Warwick, R.S., 1991, MNRAS, 248, 773 Yaqoob, T., Warwick, R.S., Pounds, K.A., 1989, MNRAS, 236, 153 Yaqoob, T., Warwick, R.S., Makino, F., Otani, C., Sokoloski, J.L., Bond, I.A., Yamauchi, M., 1993, MNRAS, 262, 435 Yaqoob, Y., Edelson, R., Weaver, K., Warwick, R.S., Mushotzky, R.F., Serlemitsos, P.J., Holt, S.S., 1995, ApJ, 453, L81 Zdziarski A. A., Johnson N. W., Done C., Smith D., McNaron-Brown K., 1995, ApJ, 438, L63 Zdziarski A. A., Johnson N. W., Magdziarz, P. 1995, MNRAS, in press Zdziarski, A.A., Magdziarz, P., 1995, MNRAS, in press
Figure captions Figure 1. The ROSAT PSPC light curves of NGC 4151 (filled circles) and the BL Lac object 1E1207.9+3945 (stars) in the low (0.1 – 0.4 keV), medium (0.5 – 1.0 keV) and high-energy (1.0 – 2.0 keV) ROSAT bands. The origin of the time axis is UT 00hr on 1993 December 1 (= Julian Date 2449322.5). Figure 2. The PSPC pulse height spectrum for NGC 4151 from the first six ROSAT observations (group A dataset) ratioed to that from the last five observations (group B dataset). Figure 3. Minimum χ2 contours for the simultaneous spectral fitting of the Group A and B spectra. The two axes correspond to fixed ratios for both the normalisation A of the hard power-law component and the column density NH across the group A and B datasets. The minimum χ2 occurs at a point consistent with a normalisation change of ∼ 2.4 but with little corresponding variation in NH . The contours correspond to 68, 90 and 99 percent confidence ranges based on a ∆χ2 criterion of 2.3, 4.6 and 9.2, respectively. Figure 4. The X-ray light curve of NGC 4151. (a) The measured fluxes in the 1–2 keV band. (b) The measured fluxes in the 2–10 keV band. The filled circles are ROSAT PSPC measurements and the diamonds are the ASCA measurements. The flux units are 10−12 erg s−1 cm2 . The horizontal dashed line corresponds to the contribution of the (non-varying) thermal bremsstrahlung component to the 1–2 keV flux. Figure 5. The ratio of the four ASCA SIS-0 spectra to the best-fitting model (the normalisation of the hard power-law is the only variable parameter across the four data sets - see text). The top to bottom panels correspond to ASCA observations 1 to 4 respectively. Figure 6. The 50-150 keV light curve from CGRO OSSE observations. Figure 7. (a) The best-fitting partial covering model (including a thermal bremsstrahlung representation of the flux below 1 keV) compared to the composite ROSAT–ASCA–OSSE count rate spectra. (b) The spectral fitting residuals below ∼ 2 keV for the ROSAT PSPC (circled crosses) and ASCA SIS 0 data (plain crosses). (c) The deconvolved 0.1 to ∼ 300 keV spectrum of NGC 4151. Figure 8. (a) A spectral model for NGC 4151 in which a warm absorber replaces the partial covering. The ionization parameter has a value ξ = 28. In this case the soft flux below 1 keV is due to a combination of scattered and leakage flux. (b) A warm absorber model with a reduced value of the ionization parameter (ξ = 14) in which the bulk of the soft flux is due to a scattered power-law component and to a soft thermal component. ˚ light curve obtained from IUE with the 1–2 keV light Figure 9. A comparison of the 1440 A curve from ROSAT and ASCA. The dashed line in the X-ray light curve corresponds to the estimated contribution of the (non-varying) soft X-ray emission (see section 3). The flux units A−1 for the ultraviolet bands and 10−12 erg cm−2 s−1 for the soft X-ray are 10−14 erg cm−2 s−1 ˚ measurements. Figure 10. A comparison of the measured absorption-corrected flux in the 2–10 keV band with A−1 in that observed at 1440 ˚ A . The flux units are 10−11 erg cm−2 s−1 and 10−14 erg cm−2 s−1 ˚ the two bands respectively. No correction for reddening has been applied to the ultraviolet data.
The dashed line represents the best-fitting linear correlation (excluding one extreme point). The dotted line is the linear correlation derived by Perola et al. (1986) from EXOSAT and IUE measurements (after scaling to the units of the plot).
Table 1: The ROSAT PSPC Observations UT Start
Nov-30 Dec-01 Dec-01 Dec-02 Dec-02 Dec-03 Dec-04 Dec-04 Dec-04 Dec-05 Dec-05 Dec-06 Dec-10 a b c
23:50:30 12:34:55 23:44:14 12:28:33 23:37:41 12:22:01 01:06:45 12:16:47 18:43:18 06:02:04 16:07:25 10:51:50 13:59:20
UT End
Dec-01 Dec-01 Dec-02 Dec-02 Dec-03 Dec-03 Dec-04 Dec-04 Dec-04 Dec-05 Dec-05 Dec-06 Dec-10
00:22:45 14:33:03 00:16:01 14:26:46 00:09:32 14:20:33 01:40:32 14:13:32 21:03:49 06:37:18 23:48:46 12:21:18 20:26:29
JD-2440000a
Exposureb
9322.50 9323.07 9323.50 9324.06 9324.50 9325.06 9325.56 9326.05 9326.33 9326.76 9327.33 9327.98 9332.22
1317 1427 1217 1412 1145 1302 1192 1387 3084 1298 6374 845 2214
Julian Date at mid-observation. Net exposure time in seconds. Observed flux in the 1–2 keV band in units of 10−12 erg cm−2 s−1 .
F1−2
c
2.3+0.6 −0.6 +0.6 3.3−0.6 +0.6 3.3−0.6 +0.6 3.1−0.6 +0.7 3.5−0.7 +0.6 2.3−0.6 +0.6 4.0−0.6 +0.6 4.3−0.6 +0.4 5.1−0.4 +0.7 6.4−0.7 +0.3 4.8−0.3 5.3+0.8 −0.8 +0.5 5.1−0.5
Table 2: The ASCA Observations UT Start
Dec-04 Dec-05 Dec-07 Dec-09 a b c d
UT End
08:33:30 21:03:46 11:14:09 20:52:25
Dec-04 Dec-06 Dec-07 Dec-10
JD-2440000a
Exposureb
9326.00 9327.49 9329.11 9331.52
10912 8392 9849 12049
15:26:34 02:36:21 18:11:51 04:07:37
F1−2
c
5.3 ± 0.2 4.9 ± 0.2 5.5 ± 0.2 4.9 ± 0.2
F2−10
23.8 ± 0.7 21.1 ± 0.7 25.1 ± 0.7 21.3 ± 0.7
Julian Date at mid-observation. Net exposure time in seconds. Observed flux in the 1–2 keV band in units of 10−12 erg cm−2 s−1 . Observed flux in the 2–10 keV band in units of 10−11 erg cm−2 s−1 .
Table 3: The CGRO OSSE Observations
UT Start
Dec-01 Dec-03 Dec-04 Dec-05 Dec-06 Dec-07 Dec-08 Dec-09 Dec-10 Dec-11 Dec-12 Dec-13 a b c
17:36:57 00:37:26 00:01:26 00:43:12 00:00:00 02:25:26 00:00:00 00:47:31 00:02:52 00:51:50 00:04:19 00:53:16
UT End
Dec-02 Dec-03 Dec-04 Dec-05 Dec-06 Dec-07 Dec-08 Dec-09 Dec-10 Dec-11 Dec-12 Dec-13
23:36:57 22:53:45 23:38:24 22:56:38 23:44:09 22:59:31 23:44:09 23:00:57 23:48:28 23:03:50 23:49:55 13:49:26
JD-2440000a
Exposureb
9323.86 9325.00 9326.00 9327.00 9328.00 9329.03 9330.00 9331.00 9332.00 9333.00 9334.00 9334.81
53490 40810 44325 41140 44700 37960 42140 41800 40440 37690 40910 23240
Julian Date at mid-observation. Net exposure time in seconds. Observed flux in the 50-150 keV band in units of 10−11 erg cm−2 s−1 .
d
F50−150
c
3.41 ± 0.23 3.98 ± 0.27 4.10 ± 0.26 4.53 ± 0.27 3.56 ± 0.26 4.17 ± 0.28 4.21 ± 0.26 4.56 ± 0.26 4.07 ± 0.27 4.83 ± 0.28 4.06 ± 0.27 4.70 ± 0.36
Table 4: Spectral fitting of the ROSAT PSPC data
a b c
Parameter
Group A/B Datasets
Group A Dataset
Group B Dataset
Aa α NH b
0.5 -
1.0+0.6 −0.3 225+70 −50
2.4+0.5 −0.4 240+30 −30
Abrem a kTbrem c NHgal b
1.4+0.3 −0.3 0.44+0.05 −0.04 2.6+0.3 −0.2
-
-
χ2 d-o-f
59.4 47
10−2 photon cm−2 s−1 keV−1 . 1020 cm−2 . keV. Table 5: The partial covering model applied to the ASCA data Parameter
a b c d
Observation 2
All
1
3
4
Aa α
0.5
6.5+0.2 −0.2 -
5.8+0.2 −0.2 -
6.9+0.2 −0.2 -
5.8+0.2 −0.2 -
NH1 d CF NH2 d
1120+370 −270 0.42+0.06 −0.05 350+30 −30
-
-
-
-
EKα c IKαb
6.34+0.03 −0.04 2.9+0.6 −0.6
-
-
-
-
Abrem a kTbrem c NHgal d
0.30+0.06 −0.05 1.3+0.3 −0.2 2.1
-
-
-
-
χ2 d-o-f
1371 1052
10−2 photon cm−2 s−1 keV−1 . 10−4 photon cm−2 s−1 . keV. 1020 cm−2 .
Table 6: The partial covering model applied to the ROSAT-ASCA-CGRO data
a b c d
Parameter
Thermal Brems. Soft Excess
Power-law Soft Excess
Aa α Ec c
6.5+0.2 −0.2 0.5 94+4 −4
6.6+0.2 −0.2 0.5 93+4 −4
NH1 d CF NH2 d
540+80 −50 0.75+0.07 −0.08 210+40 −40
630+310 −140 0.57+0.14 −0.15 320+50 −60
Abrem /Asof t b αsof t kTbrem c NHgal d
9.7+0.9 −0.8 0.51+0.03 −0.04 2.1 < 2.2
1.86+0.10 −0.11 1.3+0.2 −0.2 2.2+0.2 −0.1
χ2 d-o-f
388 300
378 300
10−2 photon cm−2 s−1 keV−1 . 10−3 photon cm−2 s−1 keV−1 . keV. 1020 cm−2 .
Table 7: The Variability Parameters Band
1–2 keV b 2–10 keV 50-150 keV
a b c
Mean Flux a
Fractional Errora
Fractional Variabilityc
0.41 22.8 4.18
0.06 0.03 0.06
0.31 0.09 0.10
Observed flux in the band in units of 10−11 erg cm−2 s−1 . Based only on the ROSAT observations. Not corrected for the statistical errors.
Figure 1:
Figure 2:
Figure 3:
Figure 4:
Figure 5:
Figure 6:
Figure 7: a - top, b - bottom, c - next page
Figure 8: a - top, b - bottom
Figure 9:
Figure 10:
